Private vs. Private proxies - 10 thread per proxy
There are several ways of getting a private proxy. The most popular will obviously be buying them from a proxy provider, but you can also set them up yourself. In the last test I compared semi-dedicated proxies versus private proxies I set up myself, but a comment made by @ron got me thinking - what if there is a difference in performance between private proxies based on if you get them from a provider or set them up yourself? This would pretty much call for a redo of the previous test. So I went for the following scenario:
a 3 hour showdown between my private proxies (called: mpp) and a batch bought off of a known proxy provider (called: bpp). 5 private proxies each, excluding recaptcha resoultion errors, how well will they perform?
Well, with mpp the average error message participation per minute was 0.94, which is more or less in line with the previous test (it was 1.09 there)
bpp however had a 1.69 participation in error message per minute which means a significantly bigger participation (an increase of 80%)
In the test the proxies behaved consistently i.e. mpp lowest fail count was 219 and highest was 286, whereas with bpp the lowest was 433 and the highest 555. The ranges did not even overlap.
Conclusion: The mpp are not on great servers, yet they still outperform bpp by a significant margin. I am in Europe, bpp were in Europe and mpp were in America - yet mpp still outperformed bpp as far as reliability is concerned. It really makes me wonder what kind of servers bpp run on and whether they are overselling.
While I am almost certain bpp would outperform semi the previous test will have to be repeated in a 3-way faceoff between bpp, mpp and semi. Based on gathered data so far i expect mpp to significantly outperform bpp and semi and bpp to outperform semi as well, just not by that much of a margin as previously tested.
What do you think? Comments, ideas? Suggestions?
a 3 hour showdown between my private proxies (called: mpp) and a batch bought off of a known proxy provider (called: bpp). 5 private proxies each, excluding recaptcha resoultion errors, how well will they perform?
Well, with mpp the average error message participation per minute was 0.94, which is more or less in line with the previous test (it was 1.09 there)
bpp however had a 1.69 participation in error message per minute which means a significantly bigger participation (an increase of 80%)
In the test the proxies behaved consistently i.e. mpp lowest fail count was 219 and highest was 286, whereas with bpp the lowest was 433 and the highest 555. The ranges did not even overlap.
Conclusion: The mpp are not on great servers, yet they still outperform bpp by a significant margin. I am in Europe, bpp were in Europe and mpp were in America - yet mpp still outperformed bpp as far as reliability is concerned. It really makes me wonder what kind of servers bpp run on and whether they are overselling.
While I am almost certain bpp would outperform semi the previous test will have to be repeated in a 3-way faceoff between bpp, mpp and semi. Based on gathered data so far i expect mpp to significantly outperform bpp and semi and bpp to outperform semi as well, just not by that much of a margin as previously tested.
What do you think? Comments, ideas? Suggestions?
Comments
As you said, i suspect the BPP will win the day.
I tested (although not as scientifically as you) 100 BPP being used on 3 servers simultaneously, versus splitting them into batches of 33, 33 and 34 and using them individually on each server.
The ones that were split seemed to perform better, even though there were roughly the same number of threads using each proxy.
Not sure why that might be, but it gives the same result that you got. Private perform better than shared.